A Meta-learning-assisted Training Framework for AI Deployment in Optical Networks
- Resource Type
- Conference
- Authors
- Liu, Xiaomin; Lun, Huazhi; Fu, Mengfan; Qiu, Qizhi; Yi, Lilin; Hu, Weisheng; Zhuge, Qunbi
- Source
- 2020 European Conference on Optical Communications (ECOC) Optical Communications (ECOC), 2020 European Conference on. :1-4 Dec, 2020
- Subject
- Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Fields, Waves and Electromagnetics
Photonics and Electrooptics
Signal Processing and Analysis
Training
Uncertain systems
Adaptation models
Simulation
Europe
Optical fiber networks
Robustness
- Language
A meta-learning-assisted training framework is proposed to improve the robustness and adaptation ability of the machine-learning-based applications deployed to practical optical networks. In simulations, applications trained by this framework can outperform the traditionally-trained one in systems with various parameter uncertainties.